Modeling the fat tails of size fluctuations in organizations

1. Mondani H, Holme P, Liljeros F (2014) Fat-Tailed Fluctuations in the Size of Organizations: The Role of Social Inﬂuence. PLoS ONE 9(7): e100527. Modeling the fat tails of size ﬂuctuations in organizations Petter Holme

2. Mondani H, Holme P, Liljeros F (2014) Fat-Tailed Fluctuations in the Size of Organizations: The Role of Social Inﬂuence. PLoS ONE 9(7): e100527. Modeling the fat tails of size ﬂuctuations in organizations Petter Holme

3. Local trade unions in Sweden, 1880–1939 -Long quiet periods -Large jumps F Liljeros, The complexity of social organizing, Ph.D. thesis 2001. Typical data: time series of sizes (not join / quit numbers) Examples

4. Local trade unions in Sweden, 1880–1939 F Liljeros, The complexity of social organizing, Ph.D. thesis 2001. Examples

5. Local trade unions in Sweden, 1880–1939 F Liljeros, The complexity of social organizing, Ph.D. thesis 2001. Examples

6. Growth rate US ﬁrms Buldyrev & al., J Phys I France 7 (1997), 635–650. Examples

7. Growth rate Italian ﬁrms Bottazzi, Secchi, Physica A 324 (2003), 213–219. Examples

8. Examples Growth rate Italian ﬁrms Bottazzi, Secchi, Physica A 324 (2003), 213–219.

9. MHRStanley&al,1996Nature379:804–806. Growthrate(someothersetof)USﬁrms Examples

10. Universality

11. Previous models

12. Previous models Economic models

13. Previous models Economic models Doesn’t ﬁt e.g. voluntary organizations

14. Physics models Previous models

15. Physics models Not without problems either… Previous models

16. Stochastic models Previous models

17. Stochastic models Previous models

18. Stochastic models Original has log- normal growth rate distribution Previous models

19. The SAF model Assumptions -N individuals connected in a network -G organizations -Each time step an agent changes organization with probability: Schwartzkopf, Axtell, Farmer, arxiv:1004.5397.

20. The SAF model Assumptions -N individuals connected in a network -G organizations -Each time step an agent changes organization with probability: Claims the network is the key (still trying just one topology)... Schwartzkopf, Axtell, Farmer, arxiv:1004.5397.

21. The SAF model Assumptions -N individuals connected in a network -G organizations -Each time step an agent changes organization with probability: Claims the network is the key (still trying just one topology)... Non-equilibrium... Schwartzkopf, Axtell, Farmer, arxiv:1004.5397.

22. The SAF model Assumptions -N individuals connected in a network -G organizations -Each time step an agent changes organization with probability: Claims the network is the key (still trying just one topology)... Non-equilibrium... Hidden parameters... Schwartzkopf, Axtell, Farmer, arxiv:1004.5397.

23. The SAF model Schwartzkopf, Axtell, Farmer, arxiv:1004.5397. cf. threshold models (Prof. Kertesz’s talk)

24. The SAF model Schwartzkopf, Axtell, Farmer, arxiv:1004.5397. cf. threshold models (Prof. Kertesz’s talk)

25. Our extended SAF model Additional assumptions -Trying diﬀerent networks -Organization cannot die (if the last person leaves a new person joins) -Attachment probability:

26. Results Tent plot, ER model δ = 1.

27. Results Tent plot, directed ER model δ = 1.

28. Results Tent plot, scale-free networks, δ = 1.

29. Results Tent plot, directed scale-free networks, δ = 1.

30. Results 2D grid, δ = 0

33. Conclusions -The SAF model works and it is independent of the network topology (it just needs a (strongly connected giant component). -The contextual inﬂuence parameter makes a diﬀerence and can cause the loss of tentity.